technical introduction to ariana rescue robot team
DESCRIPTION
This document which is presented by Amir H. Soltanzadeh outlines the technical issues applied in AriAnA rescue robot team.TRANSCRIPT
AriAnA Rescue Robot Team
Amir H. SoltanzadehRobotics Lab @ Engineering School
IAUCTB
Technical Introduction
IAUCTB 2/40Robotics Lab @ Engineering School
Outlines
Introduction to USAR Robotics• USAR as a real-world problem• RoboCup Rescue Robot League
Technical introduction• Mechanical overview• Hardware architecture• Software architecture
IAUCTB 3/40Robotics Lab @ Engineering School
USAR Robotics
IAUCTB 4/40Robotics Lab @ Engineering School
What is USAR Robotics?
USAR: Urban Search And Rescue
SearchTo look through in a place or in an area carefully in order to find something missing or lost
RescueTo free or deliver victim from confinement.
IAUCTB 5/40Robotics Lab @ Engineering School
What is USAR Robotics?
Developing robots to be used in USAR application
SearchTo look through in a place or in an area carefully in order to find something missing or lost
RescueTo free or deliver victim from confinement.
IAUCTB 6/40Robotics Lab @ Engineering School
Why use robots for USAR?
3-D law
Robots can help in Dirty, Dangerous, Dull Tasks.
They can do what rescuers or rescue dogs can’t!• voids smaller than person can enter• voids on fire or oxygen depleted
» Lose ½ cognitive attention with each level of protection
Void on fire
Void:1’x2.5’x60’
IAUCTB 7/40Robotics Lab @ Engineering School
Why use robots for USAR?
3-D law
Robots can help in Dirty, Dangerous, Dull Tasks.
The most important person in a rescue attempt is the rescuer!• Not enough trained people
» 1 survivor, entombed: 10 rescuers, 4 hours» 1 survivor, trapped/crushed: 10 rescuers, 10 hours
135 rescuers died Mexico City, 65 in confined spaces
IAUCTB 8/40Robotics Lab @ Engineering School
Why use robots for USAR?
3-D law
Robots can help in Dirty, Dangerous, Dull Tasks.
They save time!• Time is very critical
30 Min 1 Day 2 Days 3 Days 4 Days 5 Days0.0
10.0
20.0
30.0
40.0
50.0
60.0
19.4
42.2
5.61.1 0.7 0.3
1.9
9.9
9.72.2 3.0 4.0
Survival Rate
Dead
Survived
Time
% R
escu
ed
Golden 24 hours
IAUCTB 9/40Robotics Lab @ Engineering School
Taxonomy of USAR Robots
USAR robots
MAVUG
V
Man-packable
Man-portable
Big-size
USV
UAV
IAUCTB 10/40Robotics Lab @ Engineering School
Brief History of USAR Robotics
Oklahoma City bombing (1995)
The Idea of using robots in USAR domain (by R. Murphy and J. Blitch)
Hanshi-Awaji earthquake in Kobe City (1995)
The trigger for the RoboCup Rescue initiative
WTC 9/11 (2001) First practical usage of robots in real USAR application
After 2001 rescue robots were applied in several occasions:• Boat robots (USV) were used after hurricanes Charley, Dennis,
Katrina and Wilma • Aerial robots (UAV) were used after earthquake in L’Aquila, Italy
0
61 5
IAUCTB 11/40Robotics Lab @ Engineering School
RoboCup Rescue Robot League
RoboCupRoboCup
JuniorsJuniors SeniorsSeniors
SoccerSoccer RescueRescue @Home@HomeSoccerSoccer
RescueRescue
DanceDance
SimulationSimulation
Small SizeSmall Size
Middle SizeMiddle Size
Standard PlatformStandard Platform
HumanoidHumanoid
SimulationSimulation
RobotRobot
IAUCTB 12/40Robotics Lab @ Engineering School
RoboCup Rescue Robot League
Tasks• Finding victims in a simulated destructed building• Identifying detected victims (signs of life and identity)• Marking victims’ locations on an automatically generated
map
IAUCTB 13/40Robotics Lab @ Engineering School
RoboCup Rescue Robot League
Test Arena Yellow
• Ramps• Autonomous Robots
Only Orange
• Steep Ramp• Stairs
Red• Step-Field
Radio Drop-Out• Autonomous Mobility
IAUCTB 14/40Robotics Lab @ Engineering School
AriAnA Rescue Robot Team
IAUCTB 15/40Robotics Lab @ Engineering School
Brief History
Start (2005) • Research phase in Shahed Research Center (2005) Becoming official team of IAUCTB (2006) • 7th place in final ranking of RoboCup Rescue (2006) Joining with AVA – Malaysia (2008) • 2nd place in ISME 2008 student projects (2008) • 7th place in RoboCup Rescue (2009) • 1st place in Khwarizmi Robotics Competitions (2010)
2006 2007 2009 2010 2008 2009
AVA - Malaysia (ISOP Int. Co.)
IAUCTB 16/40Robotics Lab @ Engineering School
Mechanical Overview
Mobile manipulation in rough terrain: Locomotion Manipulation
IAUCTB 17/40Robotics Lab @ Engineering School
Locomotion
Mobility as a problem: Rescue robots should be highly mobile. Compromising between Mobility and Complexity of
locomotion systems is inevitable.• Biomimicry has not yet been a suitable solution due to
technical limitations:» Nature does not create efficient locomotion systems (living
beings must do numerous things).» Intelligent control of advanced mobility robots is
computationally power hungry.
Complexity
Mobility
as less complicated as possible to fulfill a task
Various Platforms)for variety of terrains(
EfficiencyComplexity
IAUCTB 18/40Robotics Lab @ Engineering School
Hybrid Locomotion
Our solution:
Legged
Designing a walking mechanism which is not necessarily inspired from the nature.
Legged systems are very hard to control!
Higher maneuverabilityon rough terrains
Triangular Tracked Wheel
Decreasing complexity of control system by means of semi-active joint controlling
WheeledTracked
Higher efficiencywhile steering
Higher traction +Lower ground pressure
IAUCTB 19/40Robotics Lab @ Engineering School
Concept of TTW Mechanism
2 DOF: Tracks (velocity & torque controlled) Triangular frames (semi-active joint):
• Active (position, velocity & torque controlled)• Passive
IAUCTB 20/40Robotics Lab @ Engineering School
Concept of TTW Mechanism
Active joint controlling:• Continuous movement: Tracks traveling → suitable for flat grounds
(This type is also available in passive mode)• Discrete movement: Triangular frames rotation → for rough terrains• Combined movement: Both tracks and triangles → for ultra-rough terrains
IAUCTB 21/40Robotics Lab @ Engineering School
Concept of TTW Mechanism
Passive joint controlling: • Surface adaptation:
» Lateral adaptation: Increasing traction without control process» Axial adaptation: Passing obstacles without control process
Not actually controlled but is monitored!
IAUCTB 22/40Robotics Lab @ Engineering School
Manipulator
Manipulator:• Surveillance
» Camera » Victim detection sensors
• Manipulation» Camera» Victim detection sensors» Gripper
Problems:• DOF:
» Maneuverability» Complexity
• Accuracy• Payload End effector’s orientation correction mechanism:
Combination of two parallelogram four-bar linkage with flexible links
IAUCTB 23/40Robotics Lab @ Engineering School
Hardware Architecture
Power Management System Main Board Communication System Motors & Drivers Video System Sensors
IAUCTB 24/40Robotics Lab @ Engineering School
Power Management System
Web based PMS:• Power distribution• Monitoring (voltage & current) • Web Interfaced• Intelligent control• Self-health check
IAUCTB 25/40Robotics Lab @ Engineering School
Main Board
Industry grade Motherboard• Small (115 x 165 mm) • Powerful
» Pentium M 1.4 GHz, 2M L2 cache• Robust
» Fanless (-40 to +80 C)» Compact Flash compatible» PC/104-plus compatible» 0% ~ 90% relative humidity
IAUCTB 26/40Robotics Lab @ Engineering School
Communications
Internal• Wired
External • Wireless Communication
» 5 GHz IEEE802.11a Access Point / Bridge
IAUCTB 27/40Robotics Lab @ Engineering School
Motors & Drivers
High efficiency brushless DC motors • ~ 90% efficient• 120 – 200W nominal power
Highly efficient Gearhead• ~ 80% efficient
Incremental Encoder• 1500 cpr
Driver• Torque control• Velocity control• Position control
IAUCTB 28/40Robotics Lab @ Engineering School
Video System
Camera• Miniature cam (QTY = 3)• Zoom cam (QTY = 1)
» Optical zoom» Auto/Manual control
Video Server• Industry grade VS
» Higher quality › Resolution: 720 x 480› Frame rate: up to 30 fps
» Robustness› 3g shock & 1g vibration
IAUCTB 29/40Robotics Lab @ Engineering School
Sensors
Navigation• Dead reckoning
» Odometry» IMU
• Range sensors» Scanning Laser Range Finder
• Vision » Monocular» Stereo
• Proximity sensors» Ultrasonic
• GPS (Outdoor only)
IAUCTB 30/40Robotics Lab @ Engineering School
Sensors
Victim identification • Temperature
» Thermal imaging camera» Temperature scanner
• Vision » Monocular
• Breathing» CO2 sensor
IAUCTB 31/40Robotics Lab @ Engineering School
Software Architecture
Robotic Server HRI SLAM
IAUCTB 32/40Robotics Lab @ Engineering School
Robotic Server
Player (started in 2000)• A universal driver for robotics
Stage• 2D multi-robot simulator
Gazebo (started in 2003)• High-fidelity 3D multi-robot simulator
IAUCTB 33/40Robotics Lab @ Engineering School
Player / Stage / Gazebo
Player)server(
Controller)client(
Controller)client(
Controller)client(
Controller)client(
Player)server(
TCP, UDP,Jini, Ice
RS232, USB, 1394, TCP, Shared Mem
Stage (2D simulation)Gazebo (3D simulation)
©Brian Gerkey
IAUCTB 34/40Robotics Lab @ Engineering School
Human Robot Interaction
Easy to understand Graphical User Interface (GUI)• Video-centric GUI
Popular X-Box controller
IAUCTB 35/40Robotics Lab @ Engineering School
SLAM
SLAM: Simultaneous Localization And Mapping• Generating a map of unknown environment while
localizing the mapping system within that map
IAUCTB 36/40Robotics Lab @ Engineering School
Navigation and SLAM
Mapping
Motion control
Localization
SLAM
Active localizationExploration
Integrated approaches
©Makarenko et al
IAUCTB 37/40Robotics Lab @ Engineering School
The SLAM Problem
Ground truth map(what happens)
Local map(what robot sees)
Global map(what robot thinks)
Given • Robot controls• Nearby measurements
Estimate• Robot state (position, orientation)• Map of world features
IAUCTB 38/40Robotics Lab @ Engineering School
Structure of SLAM Problem
uk
Xk-1
mi
mj
Xk
Zk-1,i
Zk,j
IAUCTB 39/40Robotics Lab @ Engineering School
Why SLAM is hard?
Chicken and egg problem: robot path and map are both unknown
In the real world, the mapping between observations and landmarks is unknown
Picking wrong data associations can have catastrophic consequences
Pose error correlates data associations
Robot poseuncertainty
IAUCTB 40/40Robotics Lab @ Engineering School
Questions
Thank You!